THE SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS--1990 indicates that about 20,000 murders and nonnegligent manslaughters become known to the police each year. Moreover, standardized and periodic reports indicate that a large number of these incidents remain unsolved at any time. For example, the Uniform Crime Reports [United States]: Supplementary Homicide Reports, 1976-1983 data indicate that in 1983 there were 19,653 reported homicides. The Supplementary Homicide Reports (SHR) data included 97 local law enforcement jurisdictions and were reported monthly to the FBI as part of the Uniform Crime Reporting Program. After excluding justifiable homicide and manslaughter by negligent circumstances, the SHR data indicate that out of the remaining total of 18,673 incidents, the offender is reported as unknown in 4,987 incidents, or 27.71 percent of the time. It is reasonable to assume that most police departments would prefer to report fewer unknown offenders when submitting periodic reports such as SHR's.
The 1983 SHR data indicate that the offender was reported as known in 73.29 percent of the incidents of murder and nonnegligent manslaughter and the Uniform Crime Reports--1983 indicate that the percentage of these crimes cleared by arrest during the year was 75.9 percent. The similarity between the monthly SHR and annual UCR data for 1983 suggests two things: 1) The monthly SHR's closely reflect the annual or long-run performance of police units in solving and clearing homicides. 2) If an offender is reported as known in the monthly SHR'S then the incident is cleared by an arrest.
The risk of apprehension or punishment appears to play a significant role in deterring crime as indicated by Sjoquist (1973), Phillips and Votey (1972 and 1975), Ehrlich (1973), Wilson and Boland (1978), and Howsen and Jarrell (1987). Many studies of the death penalty and homicide (Shin, 1978 provides results and authors) indicate that the risk or certainty of punishment has a consistent deterrent affect. Before particular strategies or techniques can be suggested that might increase homicide solution rates and the risk of apprehension, the nature of the problem must be clearly understood. This research is aimed at identifying the principal determinants of homicide solutions by estimating an appropriate production function and then ranking these determinants according to their relative effects. Ranking the determinants may highlight the most significant problem areas as well as indicate which programs, policies, and techniques may provide the greatest benefits. The 1983 Supplemental Homicide Reports are the most recent SHR data available and serve as the basis for the analysis.
HOMICIDE SOLUTIONS involve a production process. Since police production is not separable into different crime specific outputs (see Darrough and Heineke, 1978), police production literature must be used as the basis for model specification in the crime specific area of homicide solutions. Statistical problems would arise, however, when using cross-sectional observations of police unit inputs and homicide solution output. For example, larger police units may be more specialized. Correspondingly, their production function may be somewhat less nonseparable by crime type. Smaller and less specialized police units would likely have a greater degree of production function nonseparability. Smaller police units may also apply their resources less consistently to a given output such as homicide solutions. This is due to the different crime mixes, including fewer homicides per capita, that may occur in smaller communities. Moreover, different crime mixes indirectly cause an arbitrary degree of production separability by crime type to occur among smaller police units. This makes production comparisons between police units more difficult.
Because the SHR data used in this study provide incident level observations that are not associated with a particular police unit, the problem of production function nonseparability by crime type is avoided. …